Modelling of methane emissions utilising a Lagrangian atmospheric dispersion model in combination with Earth observation data

• Main Author: Zubas, Laimonas
• Other Authors: Monks, Paul; Boesch, Hartmut
• Published: University of Leicester 2015
• Subjects: 500
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.643683

Space-borne methane observations provide increased spatial coverage and complement the precise, but sparse network of in-situ measurement sites. In this study, a method has been developed to investigate regional-scale methane budgets using space-borne methane observations, utilising the UK Met Office Numerical Atmospheric Modelling Environment (NAME). Lagrangian atmospheric dispersion models, such as NAME, allow us to investigate fluxes at a lesser computational cost and potentially, a higher spatial resolution. An inversion algorithm was created and tested on synthetic ground measurement data. The NAME based inversion algorithm was then developed to utilise column CH4 concentrations, with an intention of applying it to Greenhouse Gases Observing SATellite (GOSAT) observations. A study utilising synthetic GOSAT-like observations was carried out, as well as synthetic inversions quantifying the performance of future methane sensing space-borne missions (CarbonSat and Sentinel-5 Precursor), when used to study fluxes over the British Isles. The results were obtained for 2 months, January and July, 2011. Sentinel-5 Precursor can reduce the flux uncertainty over England by 30% over England and Wales in July, with the remaining regions (Scotland, Republic of Ireland, Northern Ireland and northern France) achieving a reduction of 8-14%. In contrast, CarbonSat error reduction values are expected to range from 3% to 18%. Finally, we used the forward model to relate bottom-up inventories to satellite observations of atmospheric XCH4 from GOSAT. For selected regions, we have inferred patterns in atmospheric XCH4 from the spatial distribution of the surface emissions, factoring in the atmospheric transport using an atmospheric dispersion model. The forward model was found to perform poorly over Western Europe (r=0.43) and North America (r=0.48). The agreement between the observations and simulations of r=0.72 were calculated over South America, r=0.60 over South East Asia and r=0.60 over Australasia.